1 /* Extended regular expression matching and search library. 2 Copyright (C) 2002-2018 Free Software Foundation, Inc. 3 This file is part of the GNU C Library. 4 Contributed by Isamu Hasegawa <isamu@yamato.ibm.com>. 5 6 The GNU C Library is free software; you can redistribute it and/or 7 modify it under the terms of the GNU General Public 8 License as published by the Free Software Foundation; either 9 version 3 of the License, or (at your option) any later version. 10 11 The GNU C Library is distributed in the hope that it will be useful, 12 but WITHOUT ANY WARRANTY; without even the implied warranty of 13 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU 14 General Public License for more details. 15 16 You should have received a copy of the GNU General Public 17 License along with the GNU C Library; if not, see 18 <https://www.gnu.org/licenses/>. */ 19 20 #ifndef _REGEX_INTERNAL_H 21 #define _REGEX_INTERNAL_H 1 22 23 #include <assert.h> 24 #include <ctype.h> 25 #include <stdio.h> 26 #include <stdlib.h> 27 #include <string.h> 28 29 #include <langinfo.h> 30 #include <locale.h> 31 #include <wchar.h> 32 #include <wctype.h> 33 #include <stdbool.h> 34 #include <stdint.h> 35 36 #include <intprops.h> 37 38 #ifdef _LIBC 39 # include <libc-lock.h> 40 # define lock_define(name) __libc_lock_define (, name) 41 # define lock_init(lock) (__libc_lock_init (lock), 0) 42 # define lock_fini(lock) ((void) 0) 43 # define lock_lock(lock) __libc_lock_lock (lock) 44 # define lock_unlock(lock) __libc_lock_unlock (lock) 45 #elif defined GNULIB_LOCK && !defined USE_UNLOCKED_IO 46 # include "glthread/lock.h" 47 /* Use gl_lock_define if empty macro arguments are known to work. 48 Otherwise, fall back on less-portable substitutes. */ 49 # if ((defined __GNUC__ && !defined __STRICT_ANSI__) \ 50 || (defined __STDC_VERSION__ && 199901L <= __STDC_VERSION__)) 51 # define lock_define(name) gl_lock_define (, name) 52 # elif USE_POSIX_THREADS 53 # define lock_define(name) pthread_mutex_t name; 54 # elif USE_PTH_THREADS 55 # define lock_define(name) pth_mutex_t name; 56 # elif USE_SOLARIS_THREADS 57 # define lock_define(name) mutex_t name; 58 # elif USE_WINDOWS_THREADS 59 # define lock_define(name) gl_lock_t name; 60 # else 61 # define lock_define(name) 62 # endif 63 # define lock_init(lock) glthread_lock_init (&(lock)) 64 # define lock_fini(lock) glthread_lock_destroy (&(lock)) 65 # define lock_lock(lock) glthread_lock_lock (&(lock)) 66 # define lock_unlock(lock) glthread_lock_unlock (&(lock)) 67 #elif defined GNULIB_PTHREAD && !defined USE_UNLOCKED_IO 68 # include <pthread.h> 69 # define lock_define(name) pthread_mutex_t name; 70 # define lock_init(lock) pthread_mutex_init (&(lock), 0) 71 # define lock_fini(lock) pthread_mutex_destroy (&(lock)) 72 # define lock_lock(lock) pthread_mutex_lock (&(lock)) 73 # define lock_unlock(lock) pthread_mutex_unlock (&(lock)) 74 #else 75 # define lock_define(name) 76 # define lock_init(lock) 0 77 # define lock_fini(lock) ((void) 0) 78 /* The 'dfa' avoids an "unused variable 'dfa'" warning from GCC. */ 79 # define lock_lock(lock) ((void) dfa) 80 # define lock_unlock(lock) ((void) 0) 81 #endif 82 83 /* In case that the system doesn't have isblank(). */ 84 #if !defined _LIBC && ! (defined isblank || (HAVE_ISBLANK && HAVE_DECL_ISBLANK)) 85 # define isblank(ch) ((ch) == ' ' || (ch) == '\t') 86 #endif 87 88 #ifdef _LIBC 89 # ifndef _RE_DEFINE_LOCALE_FUNCTIONS 90 # define _RE_DEFINE_LOCALE_FUNCTIONS 1 91 # include <locale/localeinfo.h> 92 # include <locale/coll-lookup.h> 93 # endif 94 #endif 95 96 /* This is for other GNU distributions with internationalized messages. */ 97 #if (HAVE_LIBINTL_H && ENABLE_NLS) || defined _LIBC 98 # include <libintl.h> 99 # ifdef _LIBC 100 # undef gettext 101 # define gettext(msgid) \ 102 __dcgettext (_libc_intl_domainname, msgid, LC_MESSAGES) 103 # endif 104 #else 105 # undef gettext 106 # define gettext(msgid) (msgid) 107 #endif 108 109 #ifndef gettext_noop 110 /* This define is so xgettext can find the internationalizable 111 strings. */ 112 # define gettext_noop(String) String 113 #endif 114 115 #if (defined MB_CUR_MAX && HAVE_WCTYPE_H && HAVE_ISWCTYPE) || _LIBC 116 # define RE_ENABLE_I18N 117 #endif 118 119 /* Number of ASCII characters. */ 120 #define ASCII_CHARS 0x80 121 122 /* Number of single byte characters. */ 123 #define SBC_MAX (UCHAR_MAX + 1) 124 125 #define COLL_ELEM_LEN_MAX 8 126 127 /* The character which represents newline. */ 128 #define NEWLINE_CHAR '\n' 129 #define WIDE_NEWLINE_CHAR L'\n' 130 131 /* Rename to standard API for using out of glibc. */ 132 #ifndef _LIBC 133 # undef __wctype 134 # undef __iswalnum 135 # undef __iswctype 136 # undef __towlower 137 # undef __towupper 138 # define __wctype wctype 139 # define __iswalnum iswalnum 140 # define __iswctype iswctype 141 # define __towlower towlower 142 # define __towupper towupper 143 # define __btowc btowc 144 # define __mbrtowc mbrtowc 145 # define __wcrtomb wcrtomb 146 # define __regfree regfree 147 # define attribute_hidden 148 #endif /* not _LIBC */ 149 150 #if __GNUC__ < 3 + (__GNUC_MINOR__ < 1) 151 # define __attribute__(arg) 152 #endif 153 154 #ifndef SSIZE_MAX 155 # define SSIZE_MAX ((ssize_t) (SIZE_MAX / 2)) 156 #endif 157 158 /* The type of indexes into strings. This is signed, not size_t, 159 since the API requires indexes to fit in regoff_t anyway, and using 160 signed integers makes the code a bit smaller and presumably faster. 161 The traditional GNU regex implementation uses int for indexes. 162 The POSIX-compatible implementation uses a possibly-wider type. 163 The name 'Idx' is three letters to minimize the hassle of 164 reindenting a lot of regex code that formerly used 'int'. */ 165 typedef regoff_t Idx; 166 #ifdef _REGEX_LARGE_OFFSETS 167 # define IDX_MAX SSIZE_MAX 168 #else 169 # define IDX_MAX INT_MAX 170 #endif 171 172 /* A hash value, suitable for computing hash tables. */ 173 typedef __re_size_t re_hashval_t; 174 175 /* An integer used to represent a set of bits. It must be unsigned, 176 and must be at least as wide as unsigned int. */ 177 typedef unsigned long int bitset_word_t; 178 /* All bits set in a bitset_word_t. */ 179 #define BITSET_WORD_MAX ULONG_MAX 180 181 /* Number of bits in a bitset_word_t. For portability to hosts with 182 padding bits, do not use '(sizeof (bitset_word_t) * CHAR_BIT)'; 183 instead, deduce it directly from BITSET_WORD_MAX. Avoid 184 greater-than-32-bit integers and unconditional shifts by more than 185 31 bits, as they're not portable. */ 186 #if BITSET_WORD_MAX == 0xffffffffUL 187 # define BITSET_WORD_BITS 32 188 #elif BITSET_WORD_MAX >> 31 >> 4 == 1 189 # define BITSET_WORD_BITS 36 190 #elif BITSET_WORD_MAX >> 31 >> 16 == 1 191 # define BITSET_WORD_BITS 48 192 #elif BITSET_WORD_MAX >> 31 >> 28 == 1 193 # define BITSET_WORD_BITS 60 194 #elif BITSET_WORD_MAX >> 31 >> 31 >> 1 == 1 195 # define BITSET_WORD_BITS 64 196 #elif BITSET_WORD_MAX >> 31 >> 31 >> 9 == 1 197 # define BITSET_WORD_BITS 72 198 #elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 3 == 1 199 # define BITSET_WORD_BITS 128 200 #elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 7 == 1 201 # define BITSET_WORD_BITS 256 202 #elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 7 > 1 203 # define BITSET_WORD_BITS 257 /* any value > SBC_MAX will do here */ 204 # if BITSET_WORD_BITS <= SBC_MAX 205 # error "Invalid SBC_MAX" 206 # endif 207 #else 208 # error "Add case for new bitset_word_t size" 209 #endif 210 211 /* Number of bitset_word_t values in a bitset_t. */ 212 #define BITSET_WORDS ((SBC_MAX + BITSET_WORD_BITS - 1) / BITSET_WORD_BITS) 213 214 typedef bitset_word_t bitset_t[BITSET_WORDS]; 215 typedef bitset_word_t *re_bitset_ptr_t; 216 typedef const bitset_word_t *re_const_bitset_ptr_t; 217 218 #define PREV_WORD_CONSTRAINT 0x0001 219 #define PREV_NOTWORD_CONSTRAINT 0x0002 220 #define NEXT_WORD_CONSTRAINT 0x0004 221 #define NEXT_NOTWORD_CONSTRAINT 0x0008 222 #define PREV_NEWLINE_CONSTRAINT 0x0010 223 #define NEXT_NEWLINE_CONSTRAINT 0x0020 224 #define PREV_BEGBUF_CONSTRAINT 0x0040 225 #define NEXT_ENDBUF_CONSTRAINT 0x0080 226 #define WORD_DELIM_CONSTRAINT 0x0100 227 #define NOT_WORD_DELIM_CONSTRAINT 0x0200 228 229 typedef enum 230 { 231 INSIDE_WORD = PREV_WORD_CONSTRAINT | NEXT_WORD_CONSTRAINT, 232 WORD_FIRST = PREV_NOTWORD_CONSTRAINT | NEXT_WORD_CONSTRAINT, 233 WORD_LAST = PREV_WORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT, 234 INSIDE_NOTWORD = PREV_NOTWORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT, 235 LINE_FIRST = PREV_NEWLINE_CONSTRAINT, 236 LINE_LAST = NEXT_NEWLINE_CONSTRAINT, 237 BUF_FIRST = PREV_BEGBUF_CONSTRAINT, 238 BUF_LAST = NEXT_ENDBUF_CONSTRAINT, 239 WORD_DELIM = WORD_DELIM_CONSTRAINT, 240 NOT_WORD_DELIM = NOT_WORD_DELIM_CONSTRAINT 241 } re_context_type; 242 243 typedef struct 244 { 245 Idx alloc; 246 Idx nelem; 247 Idx *elems; 248 } re_node_set; 249 250 typedef enum 251 { 252 NON_TYPE = 0, 253 254 /* Node type, These are used by token, node, tree. */ 255 CHARACTER = 1, 256 END_OF_RE = 2, 257 SIMPLE_BRACKET = 3, 258 OP_BACK_REF = 4, 259 OP_PERIOD = 5, 260 #ifdef RE_ENABLE_I18N 261 COMPLEX_BRACKET = 6, 262 OP_UTF8_PERIOD = 7, 263 #endif /* RE_ENABLE_I18N */ 264 265 /* We define EPSILON_BIT as a macro so that OP_OPEN_SUBEXP is used 266 when the debugger shows values of this enum type. */ 267 #define EPSILON_BIT 8 268 OP_OPEN_SUBEXP = EPSILON_BIT | 0, 269 OP_CLOSE_SUBEXP = EPSILON_BIT | 1, 270 OP_ALT = EPSILON_BIT | 2, 271 OP_DUP_ASTERISK = EPSILON_BIT | 3, 272 ANCHOR = EPSILON_BIT | 4, 273 274 /* Tree type, these are used only by tree. */ 275 CONCAT = 16, 276 SUBEXP = 17, 277 278 /* Token type, these are used only by token. */ 279 OP_DUP_PLUS = 18, 280 OP_DUP_QUESTION, 281 OP_OPEN_BRACKET, 282 OP_CLOSE_BRACKET, 283 OP_CHARSET_RANGE, 284 OP_OPEN_DUP_NUM, 285 OP_CLOSE_DUP_NUM, 286 OP_NON_MATCH_LIST, 287 OP_OPEN_COLL_ELEM, 288 OP_CLOSE_COLL_ELEM, 289 OP_OPEN_EQUIV_CLASS, 290 OP_CLOSE_EQUIV_CLASS, 291 OP_OPEN_CHAR_CLASS, 292 OP_CLOSE_CHAR_CLASS, 293 OP_WORD, 294 OP_NOTWORD, 295 OP_SPACE, 296 OP_NOTSPACE, 297 BACK_SLASH 298 299 } re_token_type_t; 300 301 #ifdef RE_ENABLE_I18N 302 typedef struct 303 { 304 /* Multibyte characters. */ 305 wchar_t *mbchars; 306 307 /* Collating symbols. */ 308 # ifdef _LIBC 309 int32_t *coll_syms; 310 # endif 311 312 /* Equivalence classes. */ 313 # ifdef _LIBC 314 int32_t *equiv_classes; 315 # endif 316 317 /* Range expressions. */ 318 # ifdef _LIBC 319 uint32_t *range_starts; 320 uint32_t *range_ends; 321 # else /* not _LIBC */ 322 wchar_t *range_starts; 323 wchar_t *range_ends; 324 # endif /* not _LIBC */ 325 326 /* Character classes. */ 327 wctype_t *char_classes; 328 329 /* If this character set is the non-matching list. */ 330 unsigned int non_match : 1; 331 332 /* # of multibyte characters. */ 333 Idx nmbchars; 334 335 /* # of collating symbols. */ 336 Idx ncoll_syms; 337 338 /* # of equivalence classes. */ 339 Idx nequiv_classes; 340 341 /* # of range expressions. */ 342 Idx nranges; 343 344 /* # of character classes. */ 345 Idx nchar_classes; 346 } re_charset_t; 347 #endif /* RE_ENABLE_I18N */ 348 349 typedef struct 350 { 351 union 352 { 353 unsigned char c; /* for CHARACTER */ 354 re_bitset_ptr_t sbcset; /* for SIMPLE_BRACKET */ 355 #ifdef RE_ENABLE_I18N 356 re_charset_t *mbcset; /* for COMPLEX_BRACKET */ 357 #endif /* RE_ENABLE_I18N */ 358 Idx idx; /* for BACK_REF */ 359 re_context_type ctx_type; /* for ANCHOR */ 360 } opr; 361 #if __GNUC__ >= 2 && !defined __STRICT_ANSI__ 362 re_token_type_t type : 8; 363 #else 364 re_token_type_t type; 365 #endif 366 unsigned int constraint : 10; /* context constraint */ 367 unsigned int duplicated : 1; 368 unsigned int opt_subexp : 1; 369 #ifdef RE_ENABLE_I18N 370 unsigned int accept_mb : 1; 371 /* These 2 bits can be moved into the union if needed (e.g. if running out 372 of bits; move opr.c to opr.c.c and move the flags to opr.c.flags). */ 373 unsigned int mb_partial : 1; 374 #endif 375 unsigned int word_char : 1; 376 } re_token_t; 377 378 #define IS_EPSILON_NODE(type) ((type) & EPSILON_BIT) 379 380 struct re_string_t 381 { 382 /* Indicate the raw buffer which is the original string passed as an 383 argument of regexec(), re_search(), etc.. */ 384 const unsigned char *raw_mbs; 385 /* Store the multibyte string. In case of "case insensitive mode" like 386 REG_ICASE, upper cases of the string are stored, otherwise MBS points 387 the same address that RAW_MBS points. */ 388 unsigned char *mbs; 389 #ifdef RE_ENABLE_I18N 390 /* Store the wide character string which is corresponding to MBS. */ 391 wint_t *wcs; 392 Idx *offsets; 393 mbstate_t cur_state; 394 #endif 395 /* Index in RAW_MBS. Each character mbs[i] corresponds to 396 raw_mbs[raw_mbs_idx + i]. */ 397 Idx raw_mbs_idx; 398 /* The length of the valid characters in the buffers. */ 399 Idx valid_len; 400 /* The corresponding number of bytes in raw_mbs array. */ 401 Idx valid_raw_len; 402 /* The length of the buffers MBS and WCS. */ 403 Idx bufs_len; 404 /* The index in MBS, which is updated by re_string_fetch_byte. */ 405 Idx cur_idx; 406 /* length of RAW_MBS array. */ 407 Idx raw_len; 408 /* This is RAW_LEN - RAW_MBS_IDX + VALID_LEN - VALID_RAW_LEN. */ 409 Idx len; 410 /* End of the buffer may be shorter than its length in the cases such 411 as re_match_2, re_search_2. Then, we use STOP for end of the buffer 412 instead of LEN. */ 413 Idx raw_stop; 414 /* This is RAW_STOP - RAW_MBS_IDX adjusted through OFFSETS. */ 415 Idx stop; 416 417 /* The context of mbs[0]. We store the context independently, since 418 the context of mbs[0] may be different from raw_mbs[0], which is 419 the beginning of the input string. */ 420 unsigned int tip_context; 421 /* The translation passed as a part of an argument of re_compile_pattern. */ 422 RE_TRANSLATE_TYPE trans; 423 /* Copy of re_dfa_t's word_char. */ 424 re_const_bitset_ptr_t word_char; 425 /* true if REG_ICASE. */ 426 unsigned char icase; 427 unsigned char is_utf8; 428 unsigned char map_notascii; 429 unsigned char mbs_allocated; 430 unsigned char offsets_needed; 431 unsigned char newline_anchor; 432 unsigned char word_ops_used; 433 int mb_cur_max; 434 }; 435 typedef struct re_string_t re_string_t; 436 437 438 struct re_dfa_t; 439 typedef struct re_dfa_t re_dfa_t; 440 441 #ifndef _LIBC 442 # define IS_IN(libc) false 443 #endif 444 445 #define re_string_peek_byte(pstr, offset) \ 446 ((pstr)->mbs[(pstr)->cur_idx + offset]) 447 #define re_string_fetch_byte(pstr) \ 448 ((pstr)->mbs[(pstr)->cur_idx++]) 449 #define re_string_first_byte(pstr, idx) \ 450 ((idx) == (pstr)->valid_len || (pstr)->wcs[idx] != WEOF) 451 #define re_string_is_single_byte_char(pstr, idx) \ 452 ((pstr)->wcs[idx] != WEOF && ((pstr)->valid_len == (idx) + 1 \ 453 || (pstr)->wcs[(idx) + 1] != WEOF)) 454 #define re_string_eoi(pstr) ((pstr)->stop <= (pstr)->cur_idx) 455 #define re_string_cur_idx(pstr) ((pstr)->cur_idx) 456 #define re_string_get_buffer(pstr) ((pstr)->mbs) 457 #define re_string_length(pstr) ((pstr)->len) 458 #define re_string_byte_at(pstr,idx) ((pstr)->mbs[idx]) 459 #define re_string_skip_bytes(pstr,idx) ((pstr)->cur_idx += (idx)) 460 #define re_string_set_index(pstr,idx) ((pstr)->cur_idx = (idx)) 461 462 #if defined _LIBC || HAVE_ALLOCA 463 # include <alloca.h> 464 #endif 465 466 #ifndef _LIBC 467 # if HAVE_ALLOCA 468 /* The OS usually guarantees only one guard page at the bottom of the stack, 469 and a page size can be as small as 4096 bytes. So we cannot safely 470 allocate anything larger than 4096 bytes. Also care for the possibility 471 of a few compiler-allocated temporary stack slots. */ 472 # define __libc_use_alloca(n) ((n) < 4032) 473 # else 474 /* alloca is implemented with malloc, so just use malloc. */ 475 # define __libc_use_alloca(n) 0 476 # undef alloca 477 # define alloca(n) malloc (n) 478 # endif 479 #endif 480 481 #ifdef _LIBC 482 # define MALLOC_0_IS_NONNULL 1 483 #elif !defined MALLOC_0_IS_NONNULL 484 # define MALLOC_0_IS_NONNULL 0 485 #endif 486 487 #ifndef MAX 488 # define MAX(a,b) ((a) < (b) ? (b) : (a)) 489 #endif 490 #ifndef MIN 491 # define MIN(a,b) ((a) < (b) ? (a) : (b)) 492 #endif 493 494 #define re_malloc(t,n) ((t *) malloc ((n) * sizeof (t))) 495 #define re_realloc(p,t,n) ((t *) realloc (p, (n) * sizeof (t))) 496 #define re_free(p) free (p) 497 498 struct bin_tree_t 499 { 500 struct bin_tree_t *parent; 501 struct bin_tree_t *left; 502 struct bin_tree_t *right; 503 struct bin_tree_t *first; 504 struct bin_tree_t *next; 505 506 re_token_t token; 507 508 /* 'node_idx' is the index in dfa->nodes, if 'type' == 0. 509 Otherwise 'type' indicate the type of this node. */ 510 Idx node_idx; 511 }; 512 typedef struct bin_tree_t bin_tree_t; 513 514 #define BIN_TREE_STORAGE_SIZE \ 515 ((1024 - sizeof (void *)) / sizeof (bin_tree_t)) 516 517 struct bin_tree_storage_t 518 { 519 struct bin_tree_storage_t *next; 520 bin_tree_t data[BIN_TREE_STORAGE_SIZE]; 521 }; 522 typedef struct bin_tree_storage_t bin_tree_storage_t; 523 524 #define CONTEXT_WORD 1 525 #define CONTEXT_NEWLINE (CONTEXT_WORD << 1) 526 #define CONTEXT_BEGBUF (CONTEXT_NEWLINE << 1) 527 #define CONTEXT_ENDBUF (CONTEXT_BEGBUF << 1) 528 529 #define IS_WORD_CONTEXT(c) ((c) & CONTEXT_WORD) 530 #define IS_NEWLINE_CONTEXT(c) ((c) & CONTEXT_NEWLINE) 531 #define IS_BEGBUF_CONTEXT(c) ((c) & CONTEXT_BEGBUF) 532 #define IS_ENDBUF_CONTEXT(c) ((c) & CONTEXT_ENDBUF) 533 #define IS_ORDINARY_CONTEXT(c) ((c) == 0) 534 535 #define IS_WORD_CHAR(ch) (isalnum (ch) || (ch) == '_') 536 #define IS_NEWLINE(ch) ((ch) == NEWLINE_CHAR) 537 #define IS_WIDE_WORD_CHAR(ch) (__iswalnum (ch) || (ch) == L'_') 538 #define IS_WIDE_NEWLINE(ch) ((ch) == WIDE_NEWLINE_CHAR) 539 540 #define NOT_SATISFY_PREV_CONSTRAINT(constraint,context) \ 541 ((((constraint) & PREV_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \ 542 || ((constraint & PREV_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \ 543 || ((constraint & PREV_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context))\ 544 || ((constraint & PREV_BEGBUF_CONSTRAINT) && !IS_BEGBUF_CONTEXT (context))) 545 546 #define NOT_SATISFY_NEXT_CONSTRAINT(constraint,context) \ 547 ((((constraint) & NEXT_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \ 548 || (((constraint) & NEXT_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \ 549 || (((constraint) & NEXT_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context)) \ 550 || (((constraint) & NEXT_ENDBUF_CONSTRAINT) && !IS_ENDBUF_CONTEXT (context))) 551 552 struct re_dfastate_t 553 { 554 re_hashval_t hash; 555 re_node_set nodes; 556 re_node_set non_eps_nodes; 557 re_node_set inveclosure; 558 re_node_set *entrance_nodes; 559 struct re_dfastate_t **trtable, **word_trtable; 560 unsigned int context : 4; 561 unsigned int halt : 1; 562 /* If this state can accept "multi byte". 563 Note that we refer to multibyte characters, and multi character 564 collating elements as "multi byte". */ 565 unsigned int accept_mb : 1; 566 /* If this state has backreference node(s). */ 567 unsigned int has_backref : 1; 568 unsigned int has_constraint : 1; 569 }; 570 typedef struct re_dfastate_t re_dfastate_t; 571 572 struct re_state_table_entry 573 { 574 Idx num; 575 Idx alloc; 576 re_dfastate_t **array; 577 }; 578 579 /* Array type used in re_sub_match_last_t and re_sub_match_top_t. */ 580 581 typedef struct 582 { 583 Idx next_idx; 584 Idx alloc; 585 re_dfastate_t **array; 586 } state_array_t; 587 588 /* Store information about the node NODE whose type is OP_CLOSE_SUBEXP. */ 589 590 typedef struct 591 { 592 Idx node; 593 Idx str_idx; /* The position NODE match at. */ 594 state_array_t path; 595 } re_sub_match_last_t; 596 597 /* Store information about the node NODE whose type is OP_OPEN_SUBEXP. 598 And information about the node, whose type is OP_CLOSE_SUBEXP, 599 corresponding to NODE is stored in LASTS. */ 600 601 typedef struct 602 { 603 Idx str_idx; 604 Idx node; 605 state_array_t *path; 606 Idx alasts; /* Allocation size of LASTS. */ 607 Idx nlasts; /* The number of LASTS. */ 608 re_sub_match_last_t **lasts; 609 } re_sub_match_top_t; 610 611 struct re_backref_cache_entry 612 { 613 Idx node; 614 Idx str_idx; 615 Idx subexp_from; 616 Idx subexp_to; 617 char more; 618 char unused; 619 unsigned short int eps_reachable_subexps_map; 620 }; 621 622 typedef struct 623 { 624 /* The string object corresponding to the input string. */ 625 re_string_t input; 626 #if defined _LIBC || (defined __STDC_VERSION__ && __STDC_VERSION__ >= 199901L) 627 const re_dfa_t *const dfa; 628 #else 629 const re_dfa_t *dfa; 630 #endif 631 /* EFLAGS of the argument of regexec. */ 632 int eflags; 633 /* Where the matching ends. */ 634 Idx match_last; 635 Idx last_node; 636 /* The state log used by the matcher. */ 637 re_dfastate_t **state_log; 638 Idx state_log_top; 639 /* Back reference cache. */ 640 Idx nbkref_ents; 641 Idx abkref_ents; 642 struct re_backref_cache_entry *bkref_ents; 643 int max_mb_elem_len; 644 Idx nsub_tops; 645 Idx asub_tops; 646 re_sub_match_top_t **sub_tops; 647 } re_match_context_t; 648 649 typedef struct 650 { 651 re_dfastate_t **sifted_states; 652 re_dfastate_t **limited_states; 653 Idx last_node; 654 Idx last_str_idx; 655 re_node_set limits; 656 } re_sift_context_t; 657 658 struct re_fail_stack_ent_t 659 { 660 Idx idx; 661 Idx node; 662 regmatch_t *regs; 663 re_node_set eps_via_nodes; 664 }; 665 666 struct re_fail_stack_t 667 { 668 Idx num; 669 Idx alloc; 670 struct re_fail_stack_ent_t *stack; 671 }; 672 673 struct re_dfa_t 674 { 675 re_token_t *nodes; 676 size_t nodes_alloc; 677 size_t nodes_len; 678 Idx *nexts; 679 Idx *org_indices; 680 re_node_set *edests; 681 re_node_set *eclosures; 682 re_node_set *inveclosures; 683 struct re_state_table_entry *state_table; 684 re_dfastate_t *init_state; 685 re_dfastate_t *init_state_word; 686 re_dfastate_t *init_state_nl; 687 re_dfastate_t *init_state_begbuf; 688 bin_tree_t *str_tree; 689 bin_tree_storage_t *str_tree_storage; 690 re_bitset_ptr_t sb_char; 691 int str_tree_storage_idx; 692 693 /* number of subexpressions 're_nsub' is in regex_t. */ 694 re_hashval_t state_hash_mask; 695 Idx init_node; 696 Idx nbackref; /* The number of backreference in this dfa. */ 697 698 /* Bitmap expressing which backreference is used. */ 699 bitset_word_t used_bkref_map; 700 bitset_word_t completed_bkref_map; 701 702 unsigned int has_plural_match : 1; 703 /* If this dfa has "multibyte node", which is a backreference or 704 a node which can accept multibyte character or multi character 705 collating element. */ 706 unsigned int has_mb_node : 1; 707 unsigned int is_utf8 : 1; 708 unsigned int map_notascii : 1; 709 unsigned int word_ops_used : 1; 710 int mb_cur_max; 711 bitset_t word_char; 712 reg_syntax_t syntax; 713 Idx *subexp_map; 714 #ifdef DEBUG 715 char* re_str; 716 #endif 717 lock_define (lock) 718 }; 719 720 #define re_node_set_init_empty(set) memset (set, '\0', sizeof (re_node_set)) 721 #define re_node_set_remove(set,id) \ 722 (re_node_set_remove_at (set, re_node_set_contains (set, id) - 1)) 723 #define re_node_set_empty(p) ((p)->nelem = 0) 724 #define re_node_set_free(set) re_free ((set)->elems) 725 726 727 typedef enum 728 { 729 SB_CHAR, 730 MB_CHAR, 731 EQUIV_CLASS, 732 COLL_SYM, 733 CHAR_CLASS 734 } bracket_elem_type; 735 736 typedef struct 737 { 738 bracket_elem_type type; 739 union 740 { 741 unsigned char ch; 742 unsigned char *name; 743 wchar_t wch; 744 } opr; 745 } bracket_elem_t; 746 747 748 /* Functions for bitset_t operation. */ 749 750 static inline void 751 bitset_set (bitset_t set, Idx i) 752 { 753 set[i / BITSET_WORD_BITS] |= (bitset_word_t) 1 << i % BITSET_WORD_BITS; 754 } 755 756 static inline void 757 bitset_clear (bitset_t set, Idx i) 758 { 759 set[i / BITSET_WORD_BITS] &= ~ ((bitset_word_t) 1 << i % BITSET_WORD_BITS); 760 } 761 762 static inline bool 763 bitset_contain (const bitset_t set, Idx i) 764 { 765 return (set[i / BITSET_WORD_BITS] >> i % BITSET_WORD_BITS) & 1; 766 } 767 768 static inline void 769 bitset_empty (bitset_t set) 770 { 771 memset (set, '\0', sizeof (bitset_t)); 772 } 773 774 static inline void 775 bitset_set_all (bitset_t set) 776 { 777 memset (set, -1, sizeof (bitset_word_t) * (SBC_MAX / BITSET_WORD_BITS)); 778 if (SBC_MAX % BITSET_WORD_BITS != 0) 779 set[BITSET_WORDS - 1] = 780 ((bitset_word_t) 1 << SBC_MAX % BITSET_WORD_BITS) - 1; 781 } 782 783 static inline void 784 bitset_copy (bitset_t dest, const bitset_t src) 785 { 786 memcpy (dest, src, sizeof (bitset_t)); 787 } 788 789 static inline void 790 bitset_not (bitset_t set) 791 { 792 int bitset_i; 793 for (bitset_i = 0; bitset_i < SBC_MAX / BITSET_WORD_BITS; ++bitset_i) 794 set[bitset_i] = ~set[bitset_i]; 795 if (SBC_MAX % BITSET_WORD_BITS != 0) 796 set[BITSET_WORDS - 1] = 797 ((((bitset_word_t) 1 << SBC_MAX % BITSET_WORD_BITS) - 1) 798 & ~set[BITSET_WORDS - 1]); 799 } 800 801 static inline void 802 bitset_merge (bitset_t dest, const bitset_t src) 803 { 804 int bitset_i; 805 for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i) 806 dest[bitset_i] |= src[bitset_i]; 807 } 808 809 static inline void 810 bitset_mask (bitset_t dest, const bitset_t src) 811 { 812 int bitset_i; 813 for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i) 814 dest[bitset_i] &= src[bitset_i]; 815 } 816 817 #ifdef RE_ENABLE_I18N 818 /* Functions for re_string. */ 819 static int 820 __attribute__ ((pure, unused)) 821 re_string_char_size_at (const re_string_t *pstr, Idx idx) 822 { 823 int byte_idx; 824 if (pstr->mb_cur_max == 1) 825 return 1; 826 for (byte_idx = 1; idx + byte_idx < pstr->valid_len; ++byte_idx) 827 if (pstr->wcs[idx + byte_idx] != WEOF) 828 break; 829 return byte_idx; 830 } 831 832 static wint_t 833 __attribute__ ((pure, unused)) 834 re_string_wchar_at (const re_string_t *pstr, Idx idx) 835 { 836 if (pstr->mb_cur_max == 1) 837 return (wint_t) pstr->mbs[idx]; 838 return (wint_t) pstr->wcs[idx]; 839 } 840 841 # ifdef _LIBC 842 # include <locale/weight.h> 843 # endif 844 845 static int 846 __attribute__ ((pure, unused)) 847 re_string_elem_size_at (const re_string_t *pstr, Idx idx) 848 { 849 # ifdef _LIBC 850 const unsigned char *p, *extra; 851 const int32_t *table, *indirect; 852 uint_fast32_t nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES); 853 854 if (nrules != 0) 855 { 856 table = (const int32_t *) _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB); 857 extra = (const unsigned char *) 858 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAMB); 859 indirect = (const int32_t *) _NL_CURRENT (LC_COLLATE, 860 _NL_COLLATE_INDIRECTMB); 861 p = pstr->mbs + idx; 862 findidx (table, indirect, extra, &p, pstr->len - idx); 863 return p - pstr->mbs - idx; 864 } 865 else 866 # endif /* _LIBC */ 867 return 1; 868 } 869 #endif /* RE_ENABLE_I18N */ 870 871 #ifndef __GNUC_PREREQ 872 # if defined __GNUC__ && defined __GNUC_MINOR__ 873 # define __GNUC_PREREQ(maj, min) \ 874 ((__GNUC__ << 16) + __GNUC_MINOR__ >= ((maj) << 16) + (min)) 875 # else 876 # define __GNUC_PREREQ(maj, min) 0 877 # endif 878 #endif 879 880 #if __GNUC_PREREQ (3,4) 881 # undef __attribute_warn_unused_result__ 882 # define __attribute_warn_unused_result__ \ 883 __attribute__ ((__warn_unused_result__)) 884 #else 885 # define __attribute_warn_unused_result__ /* empty */ 886 #endif 887 888 #ifndef FALLTHROUGH 889 # if __GNUC__ < 7 890 # define FALLTHROUGH ((void) 0) 891 # else 892 # define FALLTHROUGH __attribute__ ((__fallthrough__)) 893 # endif 894 #endif 895 896 #endif /* _REGEX_INTERNAL_H */ 897